Federated Learning for Cross-Silo Electronic Health Record Prediction
Decentralized recurrent neural network predicting 30-day hospital readmission rates without sharing patient health records across hospitals.
Project Overview
Orchestrates multi-hospital collaborative AI training. Each hospital node computes local gradients on its EHR patient database and transmits encrypted weight updates to a central aggregation server, achieving high clinical accuracy while guaranteeing patient confidentiality.
Orchestrates multi-hospital collaborative AI training. Each hospital node computes local gradients on its EHR patient database and transmits encrypted weight updates to a central aggregation server, achieving high clinical accuracy while guaranteeing patient confidentiality.
Decentralized recurrent neural network predicting 30-day hospital readmission rates without sharing patient health records across hospitals.